Files
old-web/app/services/crawlerService.js
2019-07-10 12:27:30 +02:00

80 lines
2.8 KiB
JavaScript

const Promise = require("bluebird");
const OlxCrawler = require("../helpers/crawlers/olxClawler");
const db = require("../models/index");
const { allMarketAlerts } = require('../helpers/db/dbHelper');
const olxCrawler = new OlxCrawler(1, 2, 3);
const crawlers = [
olxCrawler,
];
async function crawlAll() {
console.log("CRAWLER SERVICE: crawlAll");
return Promise.map(crawlers, function (crawler) {
return crawler.crawl();
}).then(async (results) => {
try {
const marketAlertsFromDb = await allMarketAlerts(true);
const hrefs = [];
const subscribedMakretAlerts = marketAlertsFromDb.filter(marketAlert => {
return marketAlert.subscribed;
});
marketAlertsFromDb.map(marketAlert => {
if (hrefs[marketAlert.request] === undefined) {
hrefs[marketAlert.request] = []
}
hrefs[marketAlert.request].push(marketAlert.url);
})
global.hrefs = hrefs;
console.log(global.hrefs);
console.log("CRAWLER SERVICE: number of existing MarketAlerts from db: " + subscribedMakretAlerts.length);
const marketAlerts = [];
const mergedResults = [].concat.apply([], results);
for (const result of mergedResults) {
marketAlerts.push({
url: result.url,
realestateOrigin: "OLX",
originId: 1,
size: result.size,
price: result.price,
email: result.email,
request: result.uuid,
// lastDate: DataTypes.STRING,
municipality: result.municipality,
region: result.region,
gardenSize: isNaN(result.gardenSize) ? 0 : result.gardenSize,
realEstateType: result.realEstateType,
title: result.title,
notified: false
})
}
console.log("CRAWLER SERVICE: Number of crawler results: " + marketAlerts.length);
try {
const filteredMarketAlerts = marketAlerts.filter((elem) => !subscribedMakretAlerts.find(({ url }) => elem.url === url));
console.log("CRAWLER SERVICE: Number of new crawler results: " + filteredMarketAlerts.length);
await db.MarketAlert.bulkCreate(filteredMarketAlerts);
} catch (e) {
console.log("CRAWLER SERVICE: Could not bulkCreate marketalers reason: ", e);
}
} catch (e) {
console.log("CRAWLER SERVICE: Error crawling. Trying next crawler! ", e);
}
})
};
module.exports = crawlAll;
// crawlAll();